[PATCH nf 1/1] netfilter: expect: Make sure the max_expected limit is effective

[PATCH nf 1/1] netfilter: expect: Make sure the max_expected limit is effective

[gfree.wind]

From: Gao Feng <fgao@ikuai8.com>

Because the type of expecting, the member of nf_conn_help, is u8, it
would overflow after reach U8_MAX(255). So it doesn't work when we
configure the max_expected exceeds 255 with expect policy.

Now add the check for max_expected. Return the -EINVAL when it exceeds
the limit.

Signed-off-by: Gao Feng <fgao@ikuai8.com>
---
 include/net/netfilter/nf_conntrack_expect.h | 1 +
 net/netfilter/nf_conntrack_helper.c         | 3 +++
 net/netfilter/nfnetlink_cthelper.c          | 4 ++++
 3 files changed, 8 insertions(+)

diff --git a/include/net/netfilter/nf_conntrack_expect.h b/include/net/netfilter/nf_conntrack_expect.h
index 5ed33ea..aa36a31 100644
--- a/include/net/netfilter/nf_conntrack_expect.h
+++ b/include/net/netfilter/nf_conntrack_expect.h
@@ -71,6 +71,7 @@ struct nf_conntrack_expect_policy {
 };
 
 #define NF_CT_EXPECT_CLASS_DEFAULT	0
+#define NF_CT_EXPECT_MAX_CNT		U8_MAX
 
 int nf_conntrack_expect_pernet_init(struct net *net);
 void nf_conntrack_expect_pernet_fini(struct net *net);
diff --git a/net/netfilter/nf_conntrack_helper.c b/net/netfilter/nf_conntrack_helper.c
index 6dc44d9..752a977 100644
--- a/net/netfilter/nf_conntrack_helper.c
+++ b/net/netfilter/nf_conntrack_helper.c
@@ -385,6 +385,9 @@ int nf_conntrack_helper_register(struct nf_conntrack_helper *me)
 	BUG_ON(me->expect_class_max >= NF_CT_MAX_EXPECT_CLASSES);
 	BUG_ON(strlen(me->name) > NF_CT_HELPER_NAME_LEN - 1);
 
+	if (me->expect_policy->max_expected > NF_CT_EXPECT_MAX_CNT)
+		return -EINVAL;
+
 	mutex_lock(&nf_ct_helper_mutex);
 	hlist_for_each_entry(cur, &nf_ct_helper_hash[h], hnode) {
 		if (nf_ct_tuple_src_mask_cmp(&cur->tuple, &me->tuple, &mask)) {
diff --git a/net/netfilter/nfnetlink_cthelper.c b/net/netfilter/nfnetlink_cthelper.c
index 2b987d2..de841b3 100644
--- a/net/netfilter/nfnetlink_cthelper.c
+++ b/net/netfilter/nfnetlink_cthelper.c
@@ -143,6 +143,8 @@
 		nla_data(tb[NFCTH_POLICY_NAME]), NF_CT_HELPER_NAME_LEN);
 	expect_policy->max_expected =
 		ntohl(nla_get_be32(tb[NFCTH_POLICY_EXPECT_MAX]));
+	if (expect_policy->max_expected > NF_CT_EXPECT_MAX_CNT)
+		return -EINVAL;
 	expect_policy->timeout =
 		ntohl(nla_get_be32(tb[NFCTH_POLICY_EXPECT_TIMEOUT]));
 
@@ -280,6 +282,8 @@
 
 	new_policy->max_expected =
 		ntohl(nla_get_be32(tb[NFCTH_POLICY_EXPECT_MAX]));
+	if (new_policy->max_expected > NF_CT_EXPECT_MAX_CNT)
+		return -EINVAL;
 	new_policy->timeout =
 		ntohl(nla_get_be32(tb[NFCTH_POLICY_EXPECT_TIMEOUT]));
 
-- 
1.9.1

Re: [PATCH nf 1/1] netfilter: expect: Make sure the max_expected limit is effective

[Pablo Neira Ayuso]

On Fri, Mar 24, 2017 at 09:38:37AM +0800, gfree.wind@foxmail.com wrote:
> From: Gao Feng <fgao@ikuai8.com>
> 
> Because the type of expecting, the member of nf_conn_help, is u8, it
> would overflow after reach U8_MAX(255). So it doesn't work when we
> configure the max_expected exceeds 255 with expect policy.
> 
> Now add the check for max_expected. Return the -EINVAL when it exceeds
> the limit.
> 
> Signed-off-by: Gao Feng <fgao@ikuai8.com>
> ---
>  include/net/netfilter/nf_conntrack_expect.h | 1 +
>  net/netfilter/nf_conntrack_helper.c         | 3 +++
>  net/netfilter/nfnetlink_cthelper.c          | 4 ++++
>  3 files changed, 8 insertions(+)
> 
> diff --git a/include/net/netfilter/nf_conntrack_expect.h b/include/net/netfilter/nf_conntrack_expect.h
> index 5ed33ea..aa36a31 100644
> --- a/include/net/netfilter/nf_conntrack_expect.h
> +++ b/include/net/netfilter/nf_conntrack_expect.h
> @@ -71,6 +71,7 @@ struct nf_conntrack_expect_policy {
>  };
>  
>  #define NF_CT_EXPECT_CLASS_DEFAULT	0
> +#define NF_CT_EXPECT_MAX_CNT		U8_MAX

use NF_CT_EXPECT_MAX. We will expose this to userspace at some point
now that we have infrastructure to configure helpers from nft
(Florian's work already in nf-next) so use 255 instead of U8_MAX is
fine.

>  int nf_conntrack_expect_pernet_init(struct net *net);
>  void nf_conntrack_expect_pernet_fini(struct net *net);
> diff --git a/net/netfilter/nf_conntrack_helper.c b/net/netfilter/nf_conntrack_helper.c
> index 6dc44d9..752a977 100644
> --- a/net/netfilter/nf_conntrack_helper.c
> +++ b/net/netfilter/nf_conntrack_helper.c
> @@ -385,6 +385,9 @@ int nf_conntrack_helper_register(struct nf_conntrack_helper *me)
>  	BUG_ON(me->expect_class_max >= NF_CT_MAX_EXPECT_CLASSES);
>  	BUG_ON(strlen(me->name) > NF_CT_HELPER_NAME_LEN - 1);
>  
> +	if (me->expect_policy->max_expected > NF_CT_EXPECT_MAX_CNT)
> +		return -EINVAL;

I swear this is also exposed through modparams, right? So this patch
may be missing something.

RE: [PATCH nf 1/1] netfilter: expect: Make sure the max_expected limit is effective

[Gao Feng]

Hi Pablo,

> -----Original Message-----
> From: Pablo Neira Ayuso [mailto:pablo@netfilter.org]
> Sent: Friday, March 24, 2017 7:43 PM
> To: gfree.wind@foxmail.com
> Cc: netfilter-devel@vger.kernel.org; Gao Feng <fgao@ikuai8.com>
> Subject: Re: [PATCH nf 1/1] netfilter: expect: Make sure the max_expected
limit
> is effective
> 
> On Fri, Mar 24, 2017 at 09:38:37AM +0800, gfree.wind@foxmail.com wrote:
> > From: Gao Feng <fgao@ikuai8.com>
> >
> > Because the type of expecting, the member of nf_conn_help, is u8, it
> > would overflow after reach U8_MAX(255). So it doesn't work when we
> > configure the max_expected exceeds 255 with expect policy.
> >
> > Now add the check for max_expected. Return the -EINVAL when it exceeds
> > the limit.
> >
> > Signed-off-by: Gao Feng <fgao@ikuai8.com>
> > ---
> >  include/net/netfilter/nf_conntrack_expect.h | 1 +
> >  net/netfilter/nf_conntrack_helper.c         | 3 +++
> >  net/netfilter/nfnetlink_cthelper.c          | 4 ++++
> >  3 files changed, 8 insertions(+)
> >
> > diff --git a/include/net/netfilter/nf_conntrack_expect.h
> > b/include/net/netfilter/nf_conntrack_expect.h
> > index 5ed33ea..aa36a31 100644
> > --- a/include/net/netfilter/nf_conntrack_expect.h
> > +++ b/include/net/netfilter/nf_conntrack_expect.h
> > @@ -71,6 +71,7 @@ struct nf_conntrack_expect_policy {  };
> >
> >  #define NF_CT_EXPECT_CLASS_DEFAULT	0
> > +#define NF_CT_EXPECT_MAX_CNT		U8_MAX
> 
> use NF_CT_EXPECT_MAX. We will expose this to userspace at some point now
> that we have infrastructure to configure helpers from nft (Florian's work
> already in nf-next) so use 255 instead of U8_MAX is fine.
> 
> >  int nf_conntrack_expect_pernet_init(struct net *net);  void
> > nf_conntrack_expect_pernet_fini(struct net *net); diff --git
> > a/net/netfilter/nf_conntrack_helper.c
> > b/net/netfilter/nf_conntrack_helper.c
> > index 6dc44d9..752a977 100644
> > --- a/net/netfilter/nf_conntrack_helper.c
> > +++ b/net/netfilter/nf_conntrack_helper.c
> > @@ -385,6 +385,9 @@ int nf_conntrack_helper_register(struct
> nf_conntrack_helper *me)
> >  	BUG_ON(me->expect_class_max >= NF_CT_MAX_EXPECT_CLASSES);
> >  	BUG_ON(strlen(me->name) > NF_CT_HELPER_NAME_LEN - 1);
> >
> > +	if (me->expect_policy->max_expected > NF_CT_EXPECT_MAX_CNT)
> > +		return -EINVAL;
> 
> I swear this is also exposed through modparams, right? So this patch may
be
> missing something.

I thought it could be covered by nf_conntrack_helper_register, it would
return error when modparam specifies one invalid max_expected.
So I didn't check the modparam before.

Now I think you are right. It is clear to report one error by checking the
modparam.
I would send the v2 patch.
There is only one case. This is the nf_conntrack_irc.c.

Best Regards
Feng